Professional Documents
Culture Documents
(212
(
Phil Mackintosh
+ 1 212 325 5263
Phil.mackintosh@credit-suisse.com
Mark Buchanan
+ 44 212 325 4096
Mark.buchanan@credit-suisse.com
Portfolio Strate Portfolio Strategy
Why Be Average?
Market Commentary 12 May 2011
WAP Is Going on with TCA? Key Points
Shortfall has become the popular standard
to measure transaction cost, and is the
only way to measure the actual cost of
trading.
But WAP benchmarks remain popular. In
this report we find that WAP benchmarks
are far from optimal. Using modeling and
real data we show that WAPs:
Do not minimize costs
Do not minimize risk
Incentivize traders to miss opportunities
and instead be average
May not even be that useful to
measure agency costs
Shortfall Is the Only Way to Measure Cost
Implementation shortfall (IS) is well established as the industry standard
approach to measuring transaction costs. Shortfall measures the
difference between decision point (arrival price) and the execution price.
Consequently, IS quantifies the actual alpha lost due to not being able to
transact freely at the moment your decision to trade was made.
But Shortfall (alone) doesnt measure trader skill
Shortfall is just 1 data point, and it can be anything that moves a stocks
price: impact, trend, alpha decay and trader skill. So it also cant
measure trader skill on its own.
In our own TCA tool (ExPRT), we use a wide variety of metrics to help
identify trader skill, including calculating delay, trend, reversion, post-
trade alpha, spreads and interval VWAP. We then perform extensive
data bucketing to compare performance across aggression, size, sector,
start time and strategy. For more information see: ExPRT for Dummies
But WAPs Remain Popular
In a recent survey of institutional investors, Greenwich Associates found
that traditional Weighted Average Price (WAP) benchmarks continue to
be popular (exhibit 1). In fact 50% of respondents judged interval VWAP
to be extremely important or very important, and approximately 40%
felt just as strongly about PWAP.
In this report we consider: Why do traders continue to focus on WAP
benchmarks? Are WAPs good measures of transaction costs or scenario
analysis? Does targeting a particular a WAP make sense? We answer
these questions, using modeling and some real data, below.
Exhibit 1: Institutional investors increasingly
focused on shortfall instead of WAPs
0% 25% 50% 75% 100%
Order Arrival Time
Broker Placement Time
Interval VWAP
PWP
VWAP
Market on Open
Prior Nights Close
PM Decision Time
Market on Close
Extremely important Very important Neutral Somewhat important Not at all important
Based on 70 respondents who use TCA as part of their investment process
Source: Greenwich Associates
Source: Credit Suisse: AES
P
r
i
c
e
TrendScenario
Strategy
PWAP20
PWAP30
PWAP50
VWAP Is Not the Cheapest Strategy
Despite what we model in Exhibit 5, in two recent studies we found
that VWAP is not the cheapest execution strategy. In fact what is
more important is how passive you are:
- see the study below (VWAP is not the lowest cost)
- another study The Traders Trail - Reversion found that natural
block crossing had less shortfall than working similar sized orders.
VWAP is not the lowest cost
The discussions around Exhibit 5, suggest that VWAP should have the
lowest cost. However, when we looked at real trades, we found that
smarter strategies like Guerilla and Inline using the same, very low
4
aggression levels of the VWAP order actually had better performance
than VWAP orders. We also found that this occurred consistently over
time.
This implies that even if your objective is to minimize cost, you are
better off using smarter, opportunistic algorithms with their participation
turned way down instead of a systematic VWAP submission strategy.
For more information about this study and its findings see: In Search
of the "Perfect Algo".
Exhibit 7: Comparing average shortfall for
different algos over time, normalized for a very
passive basket of trades.
VIX (inverted)
INLINE
VWAP
Guerrilla
M
a
r
-
0
8
J
u
n
-
0
8
S
e
p
-
0
8
D
e
c
-
0
8
M
a
r
-
0
9
J
u
n
-
0
9
S
e
p
-
0
9
D
e
c
-
0
9
M
a
r
-
1
0
J
u
n
-
1
0
PORTFOLIO STRATEGY
Are WAPs Good Scenario Tools?
Exhibit 8: For the same sized trade, increasing
aggression increases impact
Its common practice for quant funds to back-test their
strategies using VWAP benchmarks. The key benefit of this
is that it reflects an achievable trade that also captures
alpha decay. In addition, PWAP manages opportunity cost
by automatically adjusting for unusual liquidity.
0 1 2.5 5 7.5 10 15 20 25 30 40
0
.
1
1
5 1
0 2
0
0
5
10
15
20
25
30
35
Average
Shortfall
(bps)
Aggression (%)
Trade
Size
(ADV%)
Same-Size Trades
(Shortfall as agression changes)
More and more we also hear about PWAP being used as an
execution benchmark for risk positions, especially at
aggressive PWAPs (PWAP10, 20 or 30). In theory because
these execute relatively quickly, they are achievable while also
minimizing risk and alpha decay.
Missing a Key Piece of the Puzzle
But, we also know from the work we have done on impact
cost (see: Estimating Execution Costs and Exhibit 8) that
additional trading creates impact.
When You Add trades, the WAP Moves
By extension, we can deduce that the WAP that occurred
historically would be better than the WAP if you added your
own new trades.
Source: Credit Suisse Portfolio Strategy
Exhibit 9a: Estimated stock prices with different levels o
aggression for the same order (note that the faster
execution has a shorter duration, but more impact)
f
$33.80
$33.90
$34.00
$34.10
$34.20
$34.30
$34.40
$34.50
$34.60
9:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00 14:30 15:00 15:30
PWAP20Shares
PWAP30Shares
PWAP40Shares
NaveStockPrice
PWAP20Price
PWAP30Price
PWAP40Price
We would instead propose an impact cost adjusted WAP for
scenarios. We can simulate this by adding our EDGE model
costs to an example stocks price and volumes over a day
(see Exhibit 9a). This shows that the more aggressively you
trade, the more you expect to impact the stock (see new
green, blue and red price paths). In turn this affects the
WAPs you get, as we show in Exhibit 9b:
WAPs with trading (New Model WAPs using PWAP 20,
30 and 40 prices) are higher in all instances than the
scenario WAPs (naive WAPs)
The PWAP30 (blue line) is the cheapest execution only
because it was lucky enough to execute most during
price dips. But, as the diff column in Exhibit 9 shows, it
moved the stock by more than more passive PWAP 20
and VWAP strategies
Source: Credit Suisse Portfolio Strategy
VWAP moved the new model price the least, making it
the cheapest strategy. But it was also the worst price
because it participated a lot at the afternoon highs
Source: Credit Suisse Portfolio Strategy
Exhibit 9b: Trading affects all WAPs. In our example the
only reason PWAP30 beats other PWAPs is because it
was lucky to trade the most during the price dips.
..Against you
The key takeaway here is that although you can match WAP
when you trade-along, you cant in a back-test.
Strategy NaiiveWAP NewModelWAP Diff(cps)
VWAP 34.27 34.31 0.04
PWAP20 34.21 34.28 0.06
PWAP30 34.16 34.24 0.09
PWAP40 34.16 34.29 0.13
Furthermore, any scenarios you run will understate the
transaction cost (or conversely overstate the alpha you expect
to retain).
5
PORTFOLIO STRATEGY
Does targeting a WAP make sense?
Targeting PWAP Encourages Expensive Trading
As an extension to the WAP model above, we can also show that setting a PWAP target for a trader actually creates a
significant incentive to be that PWAP, which in turn may lead to excessive costs.
In a recent report we found that a typical trade sees 5 40% reversion after a trade is finished, highest for the more
aggressive trades. (see: The Traders Trail - Reversion). When we add this reversion to the impact WAP model (from
prev page) we get price paths shown in Exhibit 10.
When set a PWAP20 target
For simplicity, we also:
Remove also trend from the model (the naive stock
price is constant (in grey) at $34.20.
Give the stock a TWAP volume curve.
Add linear impact based on executing at PWAP 15,
20 and 25.
Note that PWAP20 takes until 12:30pm. Regardless of
the actual execution strategy, the scenario benchmark
calculation will stop at 12:30pm. However the price path
for the stock varies significantly as we change the actual
execution strategy (as the blue, green and red lines show)
We find a trader is likely to see the following results:
1. Trading slower or faster introduces deviation from the
PWAP 20 benchmark, because the end time of
actual trading is different than the PWAP20
benchmark. Often this would be viewed as an
avoidable execution risk.
2. The cheapest execution is the most passive (@15%
= $34.39). Note that this is 3cents worse than
PWAP20, because the trade continues to have
impact after the PWAP20 stops calculating
3. Trading more aggressively gives you a chance to
beat PWAP20, especially if the reversion is slow.
This is because you have a lot of impact and finish
trading at the highs. Initial reversion prints are close
to these highs and keep pulling the PWAP20 price
up. However, because it is faster, it is also the most
expensive trade (@25% = $34.45).
$33.90
$34.00
$34.10
$34.20
$34.30
$34.40
$34.50
$34.60
$34.70
$34.80
9:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00 14:30 15:00 15:30
PWAP20Shares
PWAP25Shares
PWAP15Shares
NaveStockPrice
StockPriceat20%
StockPriceat25%
StockPriceat15%
PWAP20for20%
PWAP20for25%
PWAP20for15%
PWAP20
Exhibit 10: Modeling PWAP20 target vs more or less
aggressive actual trades. Note that the participation itself
changes the target PWAP, and even though reversion
happens, the WAP still rises initially as the trading finishes
at the highest point of impact
Source: Credit Suisse Portfolio Strategy
ActualAggression AvgPrice PWAP20 Diff
15% 34.39 34.35 0.03
20% 34.42 34.42
25% 34.45 34.48 (0.03)
Source: Credit Suisse Portfolio Strategy
Exhibit 11: PWAP scenarios encourage more expensive
trading strategies. The most passive execution has the
worst performance vs PWAP
From this scenario it is clear that the traders best
execution price (unless the stock has clear alpha decay or
trend) is to perfectly match the PWAP.
Based on the models and real data below, we believe targetting a WAP is
often sub-optimal as:
It encourages participation at the WAP level (so the target WAP
includes the impact that the trade will have anyway)
You discourage application of trader skill.
The WAP will understate costs if no trading occurs.
For scenario backtesting, we encourage an impact adjusted WAP, as this
will account for your impact as well as alpha decay that happens if your
signal is being used by others.
For exiting existing positions, we encourage using an accurate impact
model adjusted for estimated alpha decay (which may be based on
historic alpha progressions for that type of trade).
6
PORTFOLIO STRATEGY
Real World Results No Different
Real World Results from WAP Algos:
Exhibit 12: Deviation of Vol in Line executions
vs PWAP20, PWAP33 (below) and arrival.
Using our ExPRT TCA data, we can analyze how WAP style algos
perform in the real world. The results (in Exhibit 12) show that all
strategies miss their PWAP target.
We highlighted in Exhibit 2 that the natural randomness of prices and
volumes mean its impossible to achieve VWAP even in a perfect
scenario. But in the real world, matching a PWAP benchmark is even
more difficult because of:
7
1. Spread costs: Spread costs are significant in some countries, as
we show in Global Equity Markets Handbook. Options to
minimize spread costs include trading in dark pools and sitting at
near touch.
2. NBBO queue depth: PWAP is always computed from top of book
flow, whereas real orders usually have to wait for price and time
priority
3. Adverse selection & signaling: In order to reduce spread costs,
you need to display in lit markets. But this in turn leads to adverse
selection (you only get done straight away when the stock moves
toward you but have to chase a stock that trends away) and
signaling (if your bidding becomes persistent enough that other
participants see it). Both of which make it harder to beat the WAP.
Incentivizes Expensive Trading
Our analysis highlights that matching a WAP does not lead to transaction
cost minimization. Specifically we observe:
1. As the target participation rate increases, so does average slippage
versus arrival. Conversely, average slippage versus PWAP 33%
declines with participation rate. Versus PWAP 20%, deviation
declines as target participation rate approaches 20% and then
widens beyond 20% participation.
2. Slippage versus PWAP 20% is minimized when actual trading is
closest to 20% of the volume. Similarly, slippage versus PWAP 33%
is minimized by being closest to 33% of the volume. In other words,
trading at a rate which is different from the target PWAP
participation rate results in larger slippage versus the target PWAP.
The reasons for this are highlighted in the side box.
3. Performance versus IS declines as participation rate increases.
However, in Exhibit 12 we see that at as participation rate increases
from 5% to 20%, deviation versus PWAP 20% declines. Similarly,
as we move from 20% to 33% of the volume in Exhibit 12 deviation
versus PWAP 33% declines, and yet deviation versus IS increases.
In other words, PWAP is not a good measure of transaction cost.
Targeting a WAP benchmark incentivizes traders to trade using a WAP
strategy rather than encourage informed decision making
Source: CS Portfolio Strategy
What this Shows
Costs increase with aggression (blue line). But
trading at around 33% aggression is most likely to
get near PWAP 33 (and trading ~20% gets
nearest to PWAP 20).
This confirms that selecting an aggressive PWAP
target will increase trade costs and incent trading at
that (expensive) level of aggression.
From this analysis it is clear that minimizing slippage
versus a particular PWAP is not equivalent to
minimizing costs.
How Did We Do This?
We used our ExPRT TCA data during H2 2010.
We included only executions where the tactic
used was designed to match a WAP (i.e.
VWAP, TWAP and Volume-Inline).
We then grouped the results by realized
participation rate.
And computed the IS, PWAP 20% and PWAP
33% values for each order as a comparison.
PORTFOLIO STRATEGY
8
entation for
Disclaimer:
Please follow the attached hyperlink to an important disclosure: http://www.credit-
suisse.com/legal_terms/market_commentary_disclaimer.shtml. Structured
securities, derivatives and options are complex instruments that are not suitable for
every investor, may involve a high degree of risk, and may be appropriate
investments only for sophisticated investors who are capable of understanding and
assuming the risks involved. Supporting docum
any claims, comparisons, recommendations, statistics or other technical data will be
supplied upon request. Any trade information is preliminary
and not intended as an official transaction confirmation. Use the following links to
read the Options Clearing Corporations disclosure document:
http://www.cboe.com/LearnCenter/pdf/characteristicsandrisks.pdf
Because of the importance of tax considerations to many option transactions, the
investor considering options should consult with his/her tax advisor as to how taxes
affect the outcome of contemplated options transactions.
This material has been prepared by individual traders or sales personnel of Credit
Suisse Securities Limited and not by the Credit Suisse research department. It is
provided for informational purposes, is intended for your use only and does not
constitute an invitation or offer to subscribe for or purchase any of the products or services mentioned. The information provided is not intended to provide a sufficient basis on which
to make an investment decision. It is intended only to provide observations and views of individual traders or sales personnel, which may be different from, or inconsistent with, the
observations and views of Credit Suisse research department analysts, other Credit Suisse traders or sales personnel, or the proprietary positions of Credit Suisse. Observations and
views expressed herein may be changed by the trader or sales personnel at any time without notice. Past performance should not be taken as an indication or guarantee of future
performance, and no representation or warranty, expressed or implied is made regarding future performance. The information set forth above has been obtained from or based upon
sources believed by the trader or sales personnel to be reliable, but each of the trader or sales personnel and Credit Suisse does not represent or warrant its accuracy or
completeness and is not responsible for losses or damages arising out of errors, omissions or changes in market factors. This material does not purport to contain all of the
information that an interested party may desire and, in fact, provides only a limited view of a particular market. Credit Suisse may, from time to time, participate or invest in
transactions with issuers of securities that participate in the markets referred to herein, perform services for or solicit business from such issuers, and/or have a position or effect
transactions in the securities or derivatives thereof. The most recent Credit Suisse research on any company mentioned is at http://creditsuisse.com/researchandanalytics/
2011, CREDIT SUISSE
USA
Phil Mackintosh +1 212 325 5263 phil.mackintosh@credit-suisse.com
Victor Lin +1 212 325 5281 victor.lin@credit-suisse.com
Ana Avramovic +1 212 325 2438 ana.avramovic@credit-suisse.com
Stephen Casciano +1 212 325 0776 stephen.casciano@credit-suisse.com
Europe
Colin Goldin +44 20 7888 9637 colin.goldin@credit-suisse.com
Mark Buchanan +44 20 7888 0908 mark.k.buchanan@credit-suisse.com
Marwan Abboud +44 20 7888 0082 marwan.abboud@credit-suisse.com
Asia
Karan Karia +852 2101 6322 karan.karia@credit-suisse.com
Credit Suisse | Portfolio Strategy